METHODS ARTICLEpublished: 11 October 2013

doi: 10.3389/fneng.2013.00008

A long-lasting wireless stimulator for small mammalsIan D. Hentall*

Department of Neurological Surgery and The Miami Project to Cure Paralysis, Miller School of Medicine, University of Miami, Miami, FL, USA

Edited by:

Gabriel A. Silva, University ofCalifornia, San Diego, USA

Reviewed by:

Leandro Lorenzelli, FondazioneBruno Kessler, ItalyDomenico Caputo, La SapienzaUniversity, Italy

*Correspondence:

Ian D. Hentall, The Miami Project toCure Paralysis, R-48, PO Box016960, Miami, FL 33101-6960, USAe-mail: ihentall@med.miami.edu

The chronic effects of electrical stimulation in unrestrained awake rodents are beststudied with a wireless neural stimulator that can operate unsupervised for severalweeks or more. A robust, inexpensive, easily built, cranially implantable stimulator wasdeveloped to explore the restorative effects of brainstem stimulation after neurotrauma.Its connectorless electrodes directly protrude from a cuboid epoxy capsule containingall circuitry and power sources. This physical arrangement prevents fluid leaks or wirebreakage and also simplifies and speeds implantation. Constant-current pulses of highcompliance (34 volts) are delivered from a step-up voltage regulator under microprocessorcontrol. A slowly pulsed magnetic field controls activation state and stimulationparameters. Program status is signaled to a remote reader by interval-modulated infraredpulses. Capsule size is limited by the two batteries. Silver oxide batteries rated at8 mA-h were used routinely in 8 mm wide, 15 mm long and 4 mm high capsules.Devices of smaller contact area (5 by 12 mm) but taller (6 mm) were created for mice.Microstimulation of the rat’s raphe nuclei with intermittent 5-min (50% duty cycle)trains of 30 μA, 1 ms pulses at 8 or 24 Hz frequency during 12 daylight hours lasted21.1 days ±0.8 (mean ± standard error, Kaplan-Meir censored estimate, n = 128).Extended lifetimes (>6 weeks, no failures, n = 16) were achieved with larger batteries(44 mA-h) in longer (18 mm), taller (6 mm) capsules. The circuit and electrode designare versatile; simple modifications allowed durable constant-voltage stimulation of therat’s sciatic nerve through a cylindrical cathode from a subcutaneous pelvic capsule.Devices with these general features can address in small mammals many of the biologicaland technical questions arising neurosurgically with prolonged peripheral or deep brainstimulation.

Keywords: deep brain stimulation (DBS), rodents, chronic effects, wireless implant, brainstem

INTRODUCTIONDeep brain stimulation (DBS) is used in man to block the ongoingsymptoms of various chronic neurological disorders, most com-monly the tremor of Parkinson’s disease and related forebraindisorders (Kringelbach et al., 2010; Oluigbo et al., 2012; Lozanoand Lipsman, 2013). In a typical application, once the leads havebeen internalized successfully, stimulation is continuously appliedfor several years, which raises the possibility of inducing perma-nent changes. For example, activity-dependent axonal sproutingor synaptic potentiation may beneficially enhance the targetedpathway (Cooke and Bliss, 2006). Conversely, strong stimula-tion may produce harmful effects such as seizure foci (Albensiet al., 2007) or chemical, thermal and mechanical damage at theelectrode-tissue interface (Cogan, 2008). Studying such effectsover long periods in animal models is best done with implantablewireless stimulators. The main alternative is to tether the animalby electrically conducting cables to an external pulse generator,either continuously or for several hours daily, which risks break-age or entanglement of the intervening wires and distress to thesubject.

Recent studies in rats have shown that prolonged stimulationof brainstem raphe nuclei can enhance recovery from traumaticbrain or spinal cord injury, suggesting a treatment for early

neurotrauma in man by interim DBS (Hentall and Burns, 2009;Hentall and Gonzalez, 2012; Carballosa Gonzalez et al., 2013).The experimental evidence was acquired with a robust stimula-tion system designed to deliver patterned electrical pulse trainsto the brainstem of small mammals over many days. Full tech-nical details on this device are presented here for the first time,including justification for critical choices entailed in the design.Some readily implemented modifications, such as for peripheralnerve stimulation in rats and stimulation of the mouse’s brain,are briefly covered. Comparisons are drawn with other designspublished in recent years, illustrating a range of solutions to thetechnical challenges presented by prolonged neural stimulation insmall mammals.

METHODSCIRCUITThe schematic contains two integrated circuits (ICs): a 6-pin microprocessor (PIC10F206, SOT-23 package, MicrochipTechnology Inc.) and an 8-pin inductive step-up voltage regula-tor (LT3464, TSOT-23 package, Linear Technology Corporation).Both are of small footprint and possess the critical ability toenter a low-power “sleep” mode. Although these ICs have beenavailable for about a decade, more recent devices offer little

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 1

NEUROENGINEERING

Hentall Chronic brainstem stimulation

improvement for present purposes. Other important attributesof the 10F206 are an internal 4-MHz crystal clock, a low-powerwatchdog timer, the three output (I/O) pins and a dedicatedinput pin. Its memory is very small, consisting of 512 × 12 bitsof program memory and 24 × 8 bits of data memory, but allfunctions that are critical to wireless stimulation can be pro-grammed. Data memory was supplemented by reading or writingto unneeded control registers. Important attributes of the LT3464are a dedicated feedback pin that allows easy configuration as acurrent-source, rapid recovery from sleep and an external analoginput for optional setting of output amplitude. Its high outputcompliance (34 V) permits the use of high-impedance micro-electrodes; thus a 1 megohm microelectrode can pass 34 μA,which stimulates the typical raphe neuron within a radius ofapproximately 0.2 mm, given a threshold-distance relation of900 μA/mm2 (Hentall et al., 1984). This volume of excited tis-sue is sufficient for many experiments targeting specific nuclei insmall mammals.

A magnetic reed switch (Coto Technology, CT05-1535-J1)serves as the transducer for remote control of stimulator set-tings. It passes no current when open and, given a suitablyhigh value of in-series resistor, passes very low current whenclosed. Wireless output is signaled by an infrared (IR) light-emitting diode (Everlight Electronics Co. IR91-21C), using acustomized pulse-interval code of 3 or 4 brief (20 μs) pulses.

The current delivered by the LT3464 voltage regulator is setby a resistor (R2) attached to the FB pin (Figure 1A). Theinternal chip design provides the options of constant outputwhen the voltage at the CTRL pin is >1.25 V or variable out-put when it is <1.25 V (Shtartgot, 2003). A roughly tenfoldrange of voltages is achievable for a fixed value of R2, from34 V maximum to a minimum near the battery power supply.Most experiments using the fixed mode set the amplitude at30 μA. Experiments that called for higher currents used the vari-able mode, with output programmed in 5 steps up to 100 μA(Figure 2C). The analog control voltage (0–1 V) is created bypulse-width modulation (PWM) of one of the microproces-sor’s output pins with an RC time-constant of 0.47 ms. Digital-to-analog (DAC) conversion by PWM minimizes the requiredsize and number of components, compared with DAC or dig-ital potentiometer ICs, for example. The CTRL pin also feedsthe gate of an FET switch (Q1 in Figure 1A), which is closedby a higher voltage (>1.5 V), serving to truncate the other-wise slow output decay of the LT3464 step-up regulator aftershutdown.

MICROPROCESSOR PROGRAMS AND CONTROLPrograms are written in PICBASIC and placed in memory byMeLabs software using a U2 programmer board (all from micro-Engineering Labs Inc., Colorado Springs, CO). Magnetic input

FIGURE 1 | (A) Schematic of the stimulator circuit in the usual current-sourceconfiguration. The value for R2 was 42 kilohm for fixed amplitude stimulation(30 μA) and 10 kilohm for variable amplitude. In the alternative voltage-sourceconfiguration, R2 is 10 kilohm and an additional resistor of 40 kilohm is placedbetween the anode and cathode, giving a fixed 5-volt stimulus. (B)

Photograph of a completed, epoxy-embedded stimulator containing thelargest batteries used (type 392). Some partly visible components arelabeled. (C) A mouse brain stimulator before embedding. Thisimplementation contains the smallest batteries used (type 337). Thephotograph has the same scale as panel (B).

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 2

Hentall Chronic brainstem stimulation

FIGURE 2 | Ex vivo testing. (A) The stimulus current measured at 5different stimulus amplitudes. (B) External signaling pulses sent during thehighest amplitude current. The pattern alternated between (B1) and (B2)

during the active stimulation phase. The first pulse interval in (B1)

represents stimulus width plus clock time and the second representsstimulus amplitude; the first interval in (B2) represents pulse width and thesecond represents clock time plus amplitude. The alternation allows morethan two parameters to be encoded with two intervals, to reduce powerusage while transmitting at a reasonably fast update rate to the externalreader. Inactive phases were signaled by the presence of four pulses, sothere was no alternation of pattern. (C) Control voltages delivered duringthe stimulus pulses of graph (A). All graphs have the same time base.

wakens the microprocessor from sleep mode to begin the stimu-lation pattern, which confers a long shelf life. After activation, themain program loops continuously, emitting one stimulus pulseper loop at a frequency of 8, 16, or 24 Hz. On every 16th stimuluspulse, 3 IR signaling pulses (20 μs) are emitted in the phase ofactive stimulation and 4 pulses in the rest phase of the dutycycle (Figure 2B). During most of this loop, the microprocessorand voltage regulator are in sleep mode. The watchdog timer’speriodicity determines the stimulation frequency and clock time.

Magnetic input detected during the main cycle initiates a sub-routine that looks for additional slow magnetic pulses for the nextseveral seconds. Depending on the number of pulses received,various stimulus parameters are changed, the clock is synchro-nized to real time or the device is returned to the inactivatedstate. The stimulation pattern in most experiments was a 5-min50% duty cycle (alternating rest and stimulation) applied dailyduring 12 daylight hours. The rationale for this pattern was toavoid transmitter depletion with the 5-min rest periods and toprevent disruption of diurnal arousal cycles with the nighttimerest. The main program loop continues to signal and keep timeduring these short and long rests.

A handheld external control box detects and decodes theinfrared pulses emitted by the stimulator. Pulse width, clock time,

stimulus frequency, pulse amplitude and program status (stim-ulating or resting) are written to a liquid crystal display. Thecontrol box can optionally feed an electromagnet probe withprogrammed pulses to change the stimulator settings, althoughnormally a handheld bar magnet is more easily deployed to cre-ate pulses. The nearby presence (<2 cm) of the magnetic fieldstops the program immediately; the number of times that thefield disappears and returns within a 2 s period represents thecode. In a typical software realization, one pulse resets the clockto midday (used for synchronizing the internal clock to realtime), two pulses cycle through 5 available stimulus amplitudesor pulse widths, three pulses alternate between frequencies, 4pulses add 1 h to the clock and more than 4 pulses inactivate thedevice.

PHYSICAL CONSTRUCTION, MICROELECTRODES, AND POWER SOURCEComponents are soldered onto standard, two-sided, 0.031-thick FR4 printed circuit board (PCB) fabricated commercially.Resistors and capacitors are all surface-mount (size 0603). Thecircuit is powered by two 1.55 V silver oxide coin cells, joinedin series by a metal strip. They are attached by silver-based con-ductive epoxy (8331-14G, MG Chemicals, Surrey, BC, Canada).Both the cathodal microelectrode and the stainless steel wireanode are soldered into through-holes on the PCB. The micro-electrodes are cut to a fixed length so that they protrude fromthe epoxy capsule to reach the targeted brain region with min-imal clearance above the skull. The microelectrodes used weremade from either platinum–iridium (PI20030.5A5: Microprobe,Inc., Gaithersberg, MD) or tungsten (#573210: A-M Systems,Sequim, WA). Both types were insulated with Parylene C,rated by the manufacturers to have an “AC impedance” of 0.5megohm and roughly 0.13 mm in shaft diameter. Platinum–iridium is biologically and chemically less reactive than tung-sten, and the platinum–iridium microelectrodes never producedhistological signs of damage after 8 weeks in the rat’s mid-brain (Carballosa Gonzalez et al., 2013). The tungsten micro-electrodes are less expensive and their greater rigidity providesbetter stereotaxic accuracy in targeting deeper parts of thebrain.

For encapsulation, a low-conductance biocompatible epoxymixture (DP-270, 3 M Electronic Specialty, St. Paul, MN) isallowed to harden for 24 h or more in a suitable flexible mold.Batteries rated at 8 mA-h (at 20◦C, under specific constant load-ing) were most often used (Renata or Energizer 337; diameter4.8 mm, height 1.6 mm). For routine implantation in rats, thesewere packaged in a capsule 8 mm wide, 15 mm long, and 4 mmhigh, which weighed 1.2 ± 0.1 gm. To achieve longer lifetimes,devices with 44 mA-h batteries (Energizer 392) were fabricated(Figure 1B), necessitating longer (18 mm) and taller capsules(6 mm). For mice, a smaller cranial contact area was calledfor, so a two-board sandwiched design of 5 mm width, 12 mmlength and 6 mm height was created, containing 8 mA-h batteries(Figure 1C).

An energy budget can be calculated from manufacturer’s spec-ifications. Assuming battery power of 3 V, the microprocessordraws 240 μA when operating and 1 μA in the sleep state withan active watchdog. The step-up regulator draws 25 μA when

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 3

Hentall Chronic brainstem stimulation

operating open-circuited and 0.5 μA in the sleep state. The IRLED draws 20 mA per 20 μs pulse. From these figures, trains of 1-ms, 30-μA stimulus pulses given at 8 Hz for 12 h daily with a 50%duty cycle cause an average daily current flow of 12 μA, implyinga 56-day lifetime with two 8 mA-h batteries.

ANIMAL EXPERIMENTSAll work on animals followed protocols approved by theInstitutional Animal Care and Use Committee of the Universityof Miami, Miller School of Medicine. As described more fullyelsewhere (Hentall and Gonzalez, 2012; Carballosa Gonzalezet al., 2013), stimulators were implanted in male or femaleadult Sprague-Daley rats (250–320 g) under isoflurane anesthe-sia (1.2%) delivered by face mask, using stereotaxic guidance.The capsules were fixed by dental cement to 3 small stainlesssteel screws (00–90, 1/16th inch, Antrim Miniature Specialties,Fallbrook, CA) in the skull. The anode wire was wrapped aroundone of the screws and tied there with suture thread before thecementing. The cranial incision was left open to assist woundhealing and avoid excessive stretching of face skin around theimplant.

For peripheral nerve stimulation, the same method of anes-thesia was used for implantation and electrophysiological test-ing. Hind leg incisions were closed completely over the cap-sule by suturing; the IR pulses passed through approximately0.5 cm of tissue and were readily detectable. The cathode wasa 16-gauge ring electrode of 5 mm length that was insulatedon the outside by heat-shrink tubing, fed by an insulatedstainless steel wire. This electrode was slipped over the cutnerve. A metal eye protruded from the epoxy-embedded cap-sule to allow subdermal attachment by suturing in the hipregion. The anode, a bare stainless steel wire, was positionedsubcutaneously.

RESULTSWORKING LIFETIME OF DEVICES IN VIVOThe lifetime of activated stimulators (n = 128) with 337-type(8 mA-h) silver oxide batteries was analyzed from three seriesof experiments. All used trains of pulses (1 ms, 30 μA) thatwere alternated with 5-min rests for 12 h daily. In one series(n = 58), stimulation was applied for 7 days through platinum–iridium microelectrodes to the midbrain’s dorsal raphe or medianraphe (Carballosa Gonzalez et al., 2013). The frequency was8 Hz (n = 49) or 24 Hz (n = 9). Stimulation began 4 h (n =49) or 1 week after device implantation. The second series(n = 38) tested different durations of stimulation (usually 7or 14 days) applied through tungsten microelectrodes in thehindbrain’s nucleus raphe magnus, starting 0 or 2 or 7 daysafter implantation (Hentall and Gonzalez, 2012). In the thirdseries (not yet published), 1 week (n = 20) or 3 weeks (n =12) of stimulation was delivered through the tungsten micro-electrodes to the midbrain periaqueductal gray matter, begin-ning 0, 2, or 21 days after implantation. Electrical failures,defined as unplanned permanent cessation of the infrared sig-naling pulses, occurred in 21 of the 128 devices. Kaplan-Meirestimation using this censored data gave a mean lifetime of21.13 days (s.e.m. 0.79; estimated with SPSS version 21, IBM

Corp.). Although this lifetime is less than predicted by the batteryspecifications, high current transients and temperatures nearerto body temperature are some of the factors that may explainthe difference. There was no significant dependence on activa-tion delays or different frequencies; indeed stimulators operatingat the higher frequency (24 Hz) never failed during 7 days ofoperation. Mechanical or chemical failure of the transparentepoxy capsule or of its visible internal contents was neverobserved.

To achieve longer lifetimes, placement of batteries outside thecranially attached capsule was first explored. Lithium batteries,rated at 120 mA-h (BR-1632A; Panasonic-BSG) were implantedsubcutaneously and connected by leads to the cranially attachedstimulator in some rats (n = 6). The leads were made of insu-lated MP35N, a strong and ductile cobalt-nickel-chromium alloywith added molybdenum that has often been used for commercialpacemaker leads (Ratner et al., 1996). However, frequent break-age or disconnection of leads at their insertion points, perhapsmade more likely by the rat’s high flexibility and motility, com-pelled us to discard this approach. The desired outcome was morereadily achieved by unitary encapsulation of circuitry and batter-ies as in the original design. Using this approach, a fourth seriesof studies examined chronic PAG stimulation in rats with spinalcord injury. Active cranial stimulators with the 44 mA-h batter-ies in the larger capsules (n = 16) passed 70 μA pulses with thestandard intermittent pattern for 6 weeks without electrical orinternal mechanical failure.

OUTPUT CHARACTERISTICSIn ex vivo tests, different stimulus amplitudes were applied tomicroelectrodes in phosphate-buffered saline at 20◦C. Currentwas measured at resistor R2. Flicker can be seen superimposed onthe output (Figure 2A), due to the rapid switching of the induc-tor and the creation of control voltages by PWM (Figure 2C).These oscillations do not significantly hinder neuronal excitation,since the typical membrane time-constant is about 1 ms, and theycan be removed if desired by filtering, at the expense of slightlylonger rise times. Current density was never found to be a limitingfactor, neither in platinum–iridium nor in tungsten microelec-trodes, when stimulus amplitude ranged up to 100 μA (at 8 Hz)and frequency ranged up to 24 Hz (at 30 μA) (e.g., Figure 2A).From the area of the exposed conical tips of tungsten microelec-trodes seen under a microscope, charge density was estimated tobe 0.5 mC/cm2.

ADAPTATION FOR CONSTANT-VOLTAGE PERIPHERAL NERVESTIMULATIONThe basic circuit can readily be adapted to other uses. For exam-ple, a voltage source is preferable with gross electrodes, becauseapplied tissue potentials can be estimated and voltage is onemathematical step closer to field strength, the operative parame-ter in extracellular far field stimulation (Coburn, 1989). A voltagesource is obtained simply by placing a resistor in parallel withthe anode and cathode; for example, a 40 kilohm resistor with a10 kilohm resistor at R2 allows a 5 V stimulus.

This voltage-source circuit was used to stimulate the sciaticnerve in rats. Pulses of fixed amplitude and width (5 V, 190 μs)

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 4

Hentall Chronic brainstem stimulation

FIGURE 3 | Recordings made differentially from dorsal rootlets of

lumbar segment L4. The proximal end of the cut sciatic nerve wasstimulated at approximately 20 Hz. Each trace is from a different ratrecorded on the last day of continuous stimulation and shows tworesponses per trace. The differences between traces in the sizes andshapes of evoked response are explained by their dependence on factorssuch as the positioning of the paired recording electrodes andinter-electrode conductance.

were given for three days (n = 2) or one day (n = 2) at 20 Hz.At the end of experiments, electrical recordings were made onL4 dorsal rootlets with hook electrodes. The results showed thatthe stimulus parameters were sufficient to activate Aβ fibersbut not Aδ or C fibers (Figure 3). These recordings confirmedthe continued functional viability of the devices. The viabil-ity of the brain stimulators had been previously shown withthe histological marker 2-deoxyglucose (Carballosa Gonzalezet al., 2013). No evoked twitches were observed, but the dener-vated foot assumed a more elevated position during stimulation.These implants were physically secure in the rats for all days ofobservation.

DISCUSSIONThe range of design choices for implantable small-animal stimu-lators has been fairly stable over the last decade, given the stateof the art in electronics, batteries and electrodes. Hence, pub-lished designs and specifications can be reasonably compared.Non-commercial designs are typically optimized for quite specificexperimental applications, but sometimes can be used more gen-erally, as is the intention with commercial devices. A number ofdesigns for brain stimulation have been reported with detailedspecifications (Table 1). Similar devices have been described inless detail for long-term peripheral nerve stimulation (Al-Majedet al., 2000; Lee et al., 2010).

A major consideration for the present design was maximiza-tion of lifetime for chronic stimulation. This implies both phys-ical durability and a long-lasting power source. Durability inthe present device was achieved by one-piece encapsulation ofthe batteries, circuitry and electrode attachments within high-strength, high-resistivity epoxy. The physical integration of the

stimulator and electrodes also speeds and simplifies the surgery;the stimulator is placed on the skull and the electrode stereo-taxically positioned in the brain in one maneuver. Capsule sizeshould be minimized for the animal’s comfort and becauseoverhang or excessive material above the head risks physicaldetachment. The non-reusability of the encapsulated assemblycontaining non-rechargeable batteries was a disadvantage, butits assembly is very rapid and all components are inexpensive,electrodes being the most expensive item (unless made in thelaboratory).

A common alternative for attaining longer battery lifetimesinvolves locating some or all of the circuit components, includingbatteries or inductive receiver coils, distant from the skull, suchas subcutaneously in the back, shoulder, or abdomen, (Millardand Shepherd, 2007; Harnack et al., 2008; Arfin et al., 2009; deHaas et al., 2012). As in commercial DBS devices, subcutaneouslytunneled wires connect the electrodes with the remote compo-nents. However, tunneled leads, even when coiled, are subject torepeated bending that increases the potential for breakage andfor electrolyte penetration at connections (causing corrosion orshort circuits). In initial explorations for this project, breakagewas frequent at attachment points. Also, the wires were quicklysurrounded by new connective tissue that was quite hard andthus restricted their flexing. If leads do not break, they are pre-sumably sometimes uncomfortable for the animal. Lead breakagenear fixed implants is reported to be a common cause of failure inrigorously engineered DBS systems for humans (Alex Mohit et al.,2004; Blomstedt and Hariz, 2005), who can at least reduce the riskby voluntarily limiting movement.

Another approach to power drain is to affix temporarily a moreobtrusive device containing larger batteries on the animal’s head.A means of rapid connection is then needed between the tempo-rary part (batteries with or without electronics) and the cranialimplant (electrodes with or without electronics). Another way toextend battery lifetime is by arranging for inductive recharging,done either in situ or after temporarily detaching the device. Thismethod comes at the cost of somewhat greater electronic com-plexity and size. Alternatively, batteries can be omitted and powerbe provided by induction alone (Millard and Shepherd, 2007),which offers complete flexibility of stimulation pattern, since thisis set by the circuits driving the induction coils. A limitation isthat the external coils must be near the subject.

Remote two-way communication is very useful in wirelessstimulators. In their simplest form, input signals can turn thedevice on and off, which can save power or remove adverseeffects (de Haas et al., 2012). Output signals may simply indi-cate on or off status (Millard and Shepherd, 2007; de Haas et al.,2012). More elaborate controls can use instantaneously trans-mitted commands (Millard and Shepherd, 2007; Harnack et al.,2008) or set-and-go methods, in which parameters must be mod-ified by temporarily halting the experiment (Arfin et al., 2009;de Haas et al., 2012). In the present design, pulse width, ampli-tude, and frequency are all monitored continuously and maybe altered with a few taps of a nearby magnet during a briefperiod of intervention. An additional feature is the ability toreset the internal clock, which can be used to counter drift inthe RC watchdog timer. This is particularly useful in studies

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 5

Hentall Chronic brainstem stimulation

Table 1 | Comparison of some recent brain stimulators for small animals.

PARAMETER DEVICES

Information source Present article de Haas et al., 2012 Arfin et al., 2009 Millard andShepherd, 2007

Harnack et al., 2008

Species studied: Rat Mouse Zebra finch Rat, mouse Rat

Typical active lifetime 21 days or >42 days 10 h 12 days Indefinite 21–35 days

Battery Ag-Ag2O (2) Ag-Ag2O (3) Li (2): ML621+614 None (inductive) Ag-Ag2O (2)

Tunneled lead to head No Yes No Yes Yes

Output range 20–100 μA 20–100 μA 10–1000 μA 100–500 μA 50–600 μA

Compliance 34 V 4.65 V 5 V 5 V 12 V

Independent channels 1 2 4 1 2

Frequency 8, 16, or 24 Hz 131 Hz Large range 50–5000 Hz 131 Hz

Pulse width 100–1000 μs 60 μs, fixed 180 μs/phase, fixed 25–250μs/phase

52 μs, fixed

Biphasic No Yes Yes Yes Yes

Control medium Magnet Magnet Electromagnetic Induction coil Electromagnetic

Functions controlled Various On/off Various Various Various

Output medium Infrared Infrared Visible LED Visible LED Electromagnetic

Status readout Program variables On/off, saturated On/off On/off Program variables

Length, width, height(in mm) of main device

15, 8, 4 or 18, 8, 7 30 × 8 (diameter) 13, 13, 17* 14, 12, 16(mouse version)

38, 20, 13

Weight (g) 2.0 2.1 1.3 2.5 (mouseversion)

13

*Estimated from photographic scale.

of arousal-related areas, which need a fixed diurnal stimulationcycle. In other designs, reprogramming is done by temporarilyconnecting the device via cable to a computer, as in a commer-cial device that is detached temporarily for inductive recharging(2 Channel Wireless Stimulator, v 1.1; Triangle BioSystems, Inc.,Durham, NC, USA).

Remote communication has a significant impact on powerdrain, which depends on the medium (e.g., RF, IR, low-frequency magnetic field) and the desired range. Far-field RFconsumes a lot of power but offers long-distance multidirec-tional communication that is less affected by physical barri-ers. The pulsed directional IR beam of the presently describedstimulator was designed for minimal power drain, but it hasa short range and is highly directional. Power consumption isalso affected by the choice of stimulation parameters. Chargedelivery should be set as low as possible for the specific appli-cation. In the present device, longevity was enhanced by theuse of intermittent stimulation, although other considerationsinfluenced this choice. Thus a fairly low stimulus frequency(8 Hz) was used to match the regions’ slowly firing neurons;interference with arousal was reduced by a 12-h nocturnalpause; depletion of synaptic release or firing fatigue was han-dled by imposing 5-min gaps in a 50% duty cycle betweentrains.

Other critical trade-offs involve the number of electrodesand independent channels, which influence size, power usageand physical and electrical complexity. The present device isminimal in this respect, passing monophasic pulses through acentrally located monopolar cathode and a return wire anode.Multiple electrodes have various experimental advantages, such

as alternative targeting or synchronous stimulation of bilateralstructures, although in many experiments they are superfluous.A one-channel design can be adapted to multiple cathodes byfeeding them in parallel, but this has the drawback that therelative stimulus current is inconveniently fixed by the ratioof electrode conductances. Some devices use biphasic pulsesand some in addition offer independent control of width andamplitude for each phase (Harnack et al., 2008), which maybe useful for reducing artifact in electrical recording. However,the opposite phases do little to reverse chemical deposition orother electrochemical reactions in metal electrodes, which arevery far from equilibrium (Cogan, 2008). Additionally, closelyspaced bipolar electrodes can be used to limit the region of stim-ulation. Close spacing is essential for gross electrodes, whichcreate an approximately constant field that will cause massivenon-local stimulation if, for example, separation is several cen-timeters (not uncommon). The radial fields created by punc-tate monopolar microstimulation accomplish this localizationwith less invasiveness and without introducing the complexi-ties of anodal block (Hentall, 1987, 1989). However, the higherimpedances of microelectrodes demand a higher output compli-ance; the 34 V compliance of the present device is the highest inTable 1.

Much ingenuity continues to be given to the replace-ment of traditional electrical brain stimulation with alter-native methods of activation that are less invasive or moreselective for cell phenotype. Among interesting non-invasivesolutions, direct stimulation by ultrasound or by infraredlight can target deep regions (Bystritsky et al., 2011; Richteret al., 2011), but their transduction is of uncertain causation

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 6

Hentall Chronic brainstem stimulation

and possibly harmful to cells or tissue. External magnetic stimu-lation has frequently proved useful, although it cannot preciselyaddress discrete brain areas, especially the deeper areas (Hallett,2000). Insertion into target tissue of sub-millimeter transduc-ers that convert light or magnetic fields into electrical fields isanother intriguing possibility (Abdo et al., 2011; Kane et al.,2011; Bonmassar et al., 2012; Kim et al., 2013). Many of therecent novel approaches involve insertion of genes expressingmembrane receptors responsive to light or to chemicals thatare not normally present (Miesenbock and Kevrekidis, 2005).Nevertheless, the long-term provision of stimulation throughwearable devices to unattended wild-type animals, free to movewithin a reasonably comfortable space and unrestricted by prox-imity to a bulky apparatus, is still best done by passing cur-rent directly into inserted electrodes. The limitations of spaceand proximity imposed by battery size or power inductionwill disappear if methods to tap biological energy by electrical

conversion become practical for long-term use (Heller, 2006;Romero et al., 2009). Refinement of this methodology, exploitingadvances in the miniaturization and biocompatibility of powersources, electronics and electrodes, thus remains a productiveapproach.

ACKNOWLEDGMENTSThe work of the following is gratefully acknowledged: Scott S.Burns, Allison Irvine, and Catalina Martinez for help with thebuilding and programming of prototypes; Melissa Carballosa-Gonzalez, Joana Enes, Alberto Vitores, Lizbeth Manoah, andMaria Luisa Rodriguez for participation in animal studies; WalterJermakowicz for proofing draft versions. Funding was providedby the Robert J. Kleberg Jr. and Helen C. Kleberg Foundation, theDepartment of Defense (CDMPRP grant W81XWH-08-1-0288),the Craig H. Neilsen Foundation, the State of Florida and theMiami Project to Cure Paralysis.

REFERENCESAbdo, A., Sahin, M., Freedman, D.

S., Cevik, E., Spuhler, P. S., andUnlu, M. S. (2011). Floatinglight-activated microelectri-cal stimulators tested in therat spinal cord. J. Neural Eng.8:056012. doi: 10.1088/1741-2560/8/5/056012

Albensi, B. C., Oliver, D. R., Toupin,J., and Odero, G. (2007). Electricalstimulation protocols for hip-pocampal synaptic plasticity andneuronal hyper-excitability: are theyeffective or relevant. Exp. Neurol.204, 1–13. doi: 10.1016/j.expneurol.2006.12.009

Alex Mohit, A., Samii, A., Slimp, J.C., Grady, M. S., and Goodkin,R. (2004). Mechanical failure ofthe electrode wire in deep brainstimulation. Parkinsonism Relat.Disord. 10, 153–156. doi: 10.1016/j.parkreldis.2003.11.001

Al-Majed, A. A., Neumann, C. M.,Brushart, T. M., and Gordon,T. (2000). Brief electrical stim-ulation promotes the speedand accuracy of motor axonalregeneration. J. Neurosci. 20,2602–2608.

Arfin, S. K., Long, M. A., Fee, M.S., and Sarpeshkar, R. (2009).Wireless neural stimulation infreely behaving small animals.J. Neurophysiol. 102, 598–605. doi:10.1152/jn.00017.2009

Blomstedt, P., and Hariz, M. I. (2005).Hardware-related complicationsof deep brain stimulation: a tenyear experience. Acta Neurochir.147, 1061–1064. discussion: 4. doi:10.1007/s00701-005-0576-5

Bonmassar, G., Lee, S. W., Freeman,D. K., Polasek, M., Fried, S. I.,and Gale, J. T. (2012). Microscopic

magnetic stimulation of neural tis-sue. Nat. Commun. 3, 921. doi:10.1038/ncomms1914

Bystritsky, A., Korb, A. S., Douglas, P.K., Cohen, M. S., Melega, W. P.,Mulgaonkar, A. P., et al. (2011).A review of low-intensity focusedultrasound pulsation. Brain Stimul.4, 125–136. doi: 10.1016/j.brs.2011.03.007

Carballosa Gonzalez, M. M., Blaya, M.O., Alonso, O. F., Bramlett, H. M.,and Hentall, I. D. (2013). Midbrainraphe stimulation improves behav-ioral and anatomical recoveryfrom fluid-percussion brain injury.J. Neurotrauma 30, 119–130. doi:10.1089/neu.2012.2499

Coburn, B. (1989). Neural modelingin electrical stimulation. Crit. Rev.Biomed. Eng. 17, 133–178.

Cogan, S. F. (2008). Neural stimulationand recording electrodes. Annu.Rev. Biomed. Eng. 10, 275–309. doi:10.1146/annurev.bioeng.10.061807.160518

Cooke, S. F., and Bliss, T. V. (2006).Plasticity in the human central ner-vous system. Brain 129, 1659–1673.doi: 10.1093/brain/awl082

de Haas, R., Struikmans, R., vander Plasse, G., van Kerkhof, L.,Brakkee, J. H., Kas, M. J., et al.(2012). Wireless implantablemicro-stimulation device for highfrequency bilateral deep brainstimulation in freely moving mice.J. Neurosci. Methods 209, 113–119.doi: 10.1016/j.jneumeth.2012.05.028

Hallett, M. (2000). Transcranial mag-netic stimulation and the humanbrain. Nature 406, 147–150. doi:10.1038/35018000

Harnack, D., Meissner, W., Paulat,R., Hilgenfeld, H., Muller, W. D.,

Winter, C., et al. (2008). Continuoushigh-frequency stimulation in freelymoving rats: development of animplantable microstimulationsystem. J. Neurosci. Methods 167,278–291. doi: 10.1016/j.jneumeth.2007.08.019

Heller, A. (2006). Potentiallyimplantable miniature batter-ies. Anal. Bioanal. Chem. 385,469–473. doi: 10.1007/s00216-006-0326-4

Hentall, I. D. (1987). Practical mod-elling of monopolar axonal stim-ulation. J. Neurosci. Methods 22,65–72. doi: 10.1016/0165-0270(87)90091-4

Hentall, I. D. (1989). A theoreticalanalysis of extracellular punctatestimulation around dendrites.Neuroscience 33, 11–22. doi:10.1016/0306-4522(89)90306-0

Hentall, I. D., and Burns, S. B. (2009).Restorative effects of stimulatingmedullary raphe after spinal cordinjury. J. Rehabil. Res. Dev. 46,109–122. doi: 10.1682/JRRD.2008.04.0054

Hentall, I. D., and Gonzalez, M. M.(2012). Promotion of recoveryfrom thoracic spinal cord con-tusion in rats by stimulation ofmedullary raphe or its midbraininput. Neurorehabil. Neural Repair26, 374–384. doi: 10.1177/1545968311425178

Hentall, I. D., Zorman, G., Kansky, S.,and Fields, H. L. (1984). Relationsamong threshold, spike height,electrode distance, and conductionvelocity in electrical stimula-tion of certain medullospinalneurons. J. Neurophysiol. 51,968–977.

Kane, M. J., Breen, P. P.,Quondamatteo, F., and ÓLaighin,

G. (2011). BION microstimulators:a case study in the engineering ofan electronic implantable med-ical device. Med. Eng. Phys. 33,7–16. doi: 10.1016/j.medengphy.2010.08.010

Kim, T. I., McCall, J. G., Jung, Y. H.,Huang, X., Siuda, E. R., Li, Y., et al.(2013). Injectable, cellular-scaleoptoelectronics with applicationsfor wireless optogenetics. Science340, 211–216. doi: 10.1126/science.1232437

Kringelbach, M. L., Green, A. L.,Owen, S. L., Schweder, P. M.,and Aziz, T. Z. (2010). Sing themind electric - principles of deepbrain stimulation. Eur. J. Neurosci.32, 1070–1079. doi: 10.1111/j.1460-9568.2010.07419.x

Lee, T. H., Pan, H., Kim, I. S., Hwang, S.J., and Kim, S. J. (2010). Functionalregeneration of severed peripheralnerve using an implantable electri-cal stimulator. Conf. Proc. IEEE Eng.Med. Biol. Soc. 2010:1511–1514. doi:10.1109/IEMBS.2010.5626837

Lozano, A. M., and Lipsman, N.(2013). Probing and regulatingdysfunctional circuits using deepbrain stimulation. Neuron 77,406–424. doi: 10.1016/j.neuron.2013.01.020

Miesenbock, G., and Kevrekidis, I. G.(2005). Optical imaging and controlof genetically designated neuronsin functioning circuits. Annu.Rev. Neurosci. 28, 533–563. doi:10.1146/annurev.neuro.28.051804.101610

Millard, R. E., and Shepherd, R. K.(2007). A fully implantable stim-ulator for use in small laboratoryanimals. J. Neurosci. Methods 166,168–177. doi: 10.1016/j.jneumeth.2007.07.009

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 7

Hentall Chronic brainstem stimulation

Oluigbo, C. O., Salma, A., andRezai, A. R. (2012). Deep brainstimulation for neurological dis-orders. IEEE Rev. Biomed. Eng. 5,88–99. doi: 10.1109/RBME.2012.2197745

Ratner, B. D., Hoffman, A. S.,Schoen, F. J., and Lemons, J.E. (1996). Biomaterials Science:An Introduction to Materials inMedicine. Orlando, FL: AcademicPress.

Richter, C. P., Matic, A. I., Wells, J.D., Jansen, E. D., and Walsh, J.T. Jr. (2011). Neural stimulation

with optical radiation. LaserPhoton. Rev. 5, 68–80. doi:10.1002/lpor.200900044

Romero, E., Warrington, R. O., andNeuman, M. R. (2009). Energyscavenging sources for biomed-ical sensors. Physiol. Meas. 30,R35–R62. doi: 10.1088/0967-3334/30/9/R01

Shtartgot, L. (2003). MicropowerSOT-23 boost with integratedschottky diode provides outputdisconnect and short circuit pro-tection. Linear Technol. Mag. 13,20–24.

Conflict of Interest Statement: Theauthor declares that the researchwas conducted in the absence of anycommercial or financial relationshipsthat could be construed as a potentialconflict of interest.

Received: 27 June 2013; accepted: 17September 2013; published online: 11October 2013.Citation: Hentall ID (2013) A long-lasting wireless stimulator for smallmammals. Front. Neuroeng. 6:8. doi:10.3389/fneng.2013.00008

This article was submitted to the journalFrontiers in Neuroengineering.Copyright © 2013 Hentall. This isan open-access article distributed underthe terms of the Creative CommonsAttribution License (CC BY). The use,distribution or reproduction in otherforums is permitted, provided the orig-inal author(s) or licensor are cred-ited and that the original publicationin this journal is cited, in accordancewith accepted academic practice. No use,distribution or reproduction is permit-ted which does not comply with theseterms.

Frontiers in Neuroengineering www.frontiersin.org October 2013 | Volume 6 | Article 8 | 8